create src/predict_drug_target folder for importing from other projects

.github/workflows/test.yml

@@ -42,7 +42,7 @@ jobs:
     - name: Prepare data and train the model
       run: |
         ./prepare.sh
-        python src/train.py
+        python src/predict_drug_target/train.py
 
     - name: Run tests
       run: |

Dockerfile

@@ -25,4 +25,4 @@ RUN pip3 install -r requirements.txt
 ADD . .
 RUN pip3 install -e .
 
-CMD [ "gunicorn", "-k", "uvicorn.workers.UvicornWorker", "-b", "0.0.0.0:8000", "--workers", "4", "src.api:app" ]
+CMD [ "gunicorn", "-k", "uvicorn.workers.UvicornWorker", "-b", "0.0.0.0:8000", "--workers", "4", "src.predict_drug_target.api:app" ]

README.md

@@ -42,7 +42,7 @@ Query the Bio2RDF endpoint to get drugs and their smiles, targets and their prot
 
 Process the Bio2RDF data to generate the inputs needed for the two embeddings methods
 ```bash
-python src/prepare.py
+python src/predict_drug_target/prepare.py
 ```
 
 Install the ESM library
@@ -84,7 +84,7 @@ export CUDA_VISIBLE_DEVICES=1
 Train the model:
 
 ```bash
-python src/train.py
+python src/predict_drug_target/train.py
 ```
 
 > Results are in the `results/` folder, model pickle goes to the `models/` folder
@@ -94,7 +94,7 @@ python src/train.py
 Run the prediction workflow for 2 entities:
 
 ```bash
-python src/predict.py
+python src/predict_drug_target/predict.py
 ```
 
 Users provides drugs and targets using their CHEMBL or Ensembl IDs, the script will test all provided drugs against all provided targets, and return a prediction score (how confident we are that the drug interacts with the target) for each drug-target pair.

archive/old_train.py

@@ -11,9 +11,9 @@
 from sklearn import ensemble, metrics
 from sklearn.model_selection import StratifiedKFold
 
-from src.embeddings import compute_drug_embedding, compute_target_embedding
-from src.utils import COLLECTIONS, log
-from src.vectordb import init_vectordb
+from predict_drug_target.embeddings import compute_drug_embedding, compute_target_embedding
+from predict_drug_target.utils import COLLECTIONS, log
+from predict_drug_target.vectordb import init_vectordb
 
 vectordb = init_vectordb(recreate=False)
 

archive/prepare_opentargets.py

@@ -6,8 +6,8 @@
 import pandas as pd
 from tqdm import tqdm
 
-from src.utils import ACCEPTED_NAMESPACES, COLLECTIONS, get_pref_ids, get_seq_for_target, get_smiles_for_drug, log
-from src.vectordb import init_vectordb
+from predict_drug_target.utils import ACCEPTED_NAMESPACES, COLLECTIONS, get_pref_ids, get_seq_for_target, get_smiles_for_drug, log
+from predict_drug_target.vectordb import init_vectordb
 
 # NOTE: Download opentargets before running this script
 # ./scripts/download_opentargets.sh

docker-compose.yml

@@ -24,7 +24,7 @@ services:
     shm_size: '4g'
     # ports:
     #   - 8000:8000
-    command: uvicorn --host 0.0.0.0 src.api:app
+    command: uvicorn --host 0.0.0.0 src.predict_drug_target.api:app
     # --reload
     networks:
       - nginx

notebooks/get_similarities.ipynb

@@ -23,7 +23,7 @@
     }
    ],
    "source": [
-    "from src.vectordb import init_vectordb\n",
+    "from predict_drug_target.vectordb import init_vectordb\n",
     "\n",
     "vectordb = init_vectordb(recreate=False)"
    ]

prepare.sh

@@ -33,4 +33,4 @@ cd ..
 
 
 echo "Generate list of known_drug_target pairs for OpenTargets"
-python3 src/prepare.py
+python3 src/predict_drug_target/prepare.py

pyproject.toml

@@ -69,7 +69,7 @@ test = [
 ]
 
 [project.scripts]
-predict-dt = "src.__main__:cli"
+predict-dt = "src.predict_drug_target.__main__:cli"
 
 
 [project.urls]
@@ -87,9 +87,9 @@ features = [
 ]
 
 [tool.hatch.envs.default.scripts]
-train = "python3 src/train.py {args}"
-predict = "python3 src/predict.py {args}"
-api = "uvicorn src.api:app --host 0.0.0.0 --reload {args}"
+train = "python3 src/predict_drug_target/train.py {args}"
+predict = "python3 src/predict_drug_target/predict.py {args}"
+api = "uvicorn src.predict_drug_target.api:app --host 0.0.0.0 --reload {args}"
 fmt = [
     "black src/",
     "ruff src/ --fix",
@@ -109,7 +109,7 @@ requirements = "pip-compile -o requirements.txt pyproject.toml"
 
 # TOOLS
 [tool.hatch.build.targets.wheel]
-packages = ["src"]
+packages = ["src/predict_drug_target"]
 
 [tool.hatch.metadata]
 allow-direct-references = true

src/__main__.py → src/predict_drug_target/__main__.py

@@ -1,6 +1,6 @@
 import typer
 
-# from src.embeddings import compute
+# from predict_drug_target.embeddings import compute
 
 cli = typer.Typer()
 

src/api.py → src/predict_drug_target/api.py

@@ -3,10 +3,10 @@
 
 from trapi_predict_kit import TRAPI, settings
 
-from src.predict import get_drug_target_predictions
-from src.train import train
-from src.utils import COLLECTIONS
-from src.vectordb import init_vectordb
+from predict_drug_target.predict import get_drug_target_predictions
+from predict_drug_target.train import train
+from predict_drug_target.utils import COLLECTIONS
+from predict_drug_target.vectordb import init_vectordb
 
 log_level = logging.INFO
 logging.basicConfig(level=log_level)

src/embeddings.py → src/predict_drug_target/embeddings.py

@@ -8,7 +8,7 @@
 from smiles_transformer import get_smiles_embeddings
 from tqdm import tqdm
 
-from src.utils import (
+from predict_drug_target.utils import (
     ACCEPTED_NAMESPACES,
     EMBEDDINGS_SIZE_DRUG,
     EMBEDDINGS_SIZE_TARGET,
@@ -18,7 +18,7 @@
     get_smiles_for_drug,
     log,
 )
-from src.vectordb import VectorDB, init_vectordb
+from predict_drug_target.vectordb import VectorDB, init_vectordb
 
 VECTORDB = init_vectordb(recreate=False)
 

src/predict.py → src/predict_drug_target/predict.py

@@ -4,14 +4,14 @@
 import pandas as pd
 from trapi_predict_kit import PredictInput, PredictOutput, trapi_predict
 
-from src.embeddings import compute_drug_embedding, compute_target_embedding
-from src.utils import (
+from predict_drug_target.embeddings import compute_drug_embedding, compute_target_embedding
+from predict_drug_target.utils import (
     BOLD,
     COLLECTIONS,
     END,
     log,
 )
-from src.vectordb import init_vectordb
+from predict_drug_target.vectordb import init_vectordb
 
 VECTORDB = init_vectordb(recreate=False)
 

src/prepare.py → src/predict_drug_target/prepare.py

@@ -6,10 +6,10 @@
 from tqdm import tqdm
 from src import vectordb
 
-# from src.embeddings import compute_drug_embedding, compute_target_embedding
-from src.embeddings import compute
-from src.utils import COLLECTIONS, log, get_pref_ids
-from src.vectordb import init_vectordb
+# from predict_drug_target.embeddings import compute_drug_embedding, compute_target_embedding
+from predict_drug_target.embeddings import compute
+from predict_drug_target.utils import COLLECTIONS, log, get_pref_ids
+from predict_drug_target.vectordb import init_vectordb
 
 # NOTE: script to run the WHOLE pipeline on opentargets data
 # it will automatically compute embeddings for all drugs and targets

src/train.py → src/predict_drug_target/train.py

@@ -19,8 +19,8 @@
 import xgboost as xgb
 from xgboost import XGBClassifier, DMatrix
 
-from src.utils import log, TrainingConfig
-from src.vectordb import init_vectordb
+from predict_drug_target.utils import log, TrainingConfig
+from predict_drug_target.vectordb import init_vectordb
 
 vectordb = init_vectordb(recreate=False)
 

src/vectordb.py → src/predict_drug_target/vectordb.py

@@ -13,7 +13,7 @@
     SearchParams,
 )
 
-from src.utils import log, COLLECTIONS
+from predict_drug_target.utils import log, COLLECTIONS
 
 
 # Define an abstract class VectorDB

tests/test_trapi.py

@@ -5,7 +5,7 @@
 from trapi_predict_kit import settings
 from reasoner_validator.validator import TRAPIResponseValidator
 
-from src.api import app, trapi_example
+from predict_drug_target.api import app, trapi_example
 
 
 client = TestClient(app)